AI Gateway begins supporting Regional Inference

You can now use Regional Inference through AI Gateway to pin requests to United States (US) or European (EU) data centers. Using the inferenceRegion setting on a request, inference runs within the selected region, and any data retained by the corresponding model provider is also stored in that region.

The supported routing options are as follows:

• us: US data centers
• eu: EU data centers
• global: All regions (default)

If no model provider capable of handling the request is available in the selected region, the request is designed to fail rather than being rerouted to another region. In addition, every response includes the region where the inference actually took place, so you can verify data residency directly.

Previously, meeting data residency or compliance requirements required setting up separate routing configurations for each provider, but now a single field provides a unified experience that works the same way everywhere. Note that pinning to a region may incur regional rates roughly 10% higher than standard pricing, depending on the model provider.